>>On 2012-12-19 21:06, Neil Horman wrote: >> ... >> +.BR \-u\ <pid> , " \-\-uts=<pid>" >> +Migrate to the uts namespace of <pid> >> +.TP >> +.BR \-i\ <pid> , " \-\-ipc=<pid>" >> +Migrate to the ipc namespace of <pid> >> +.TP >> +.BR \-n\ <pid> , " \-\-net\<pid>" >> +Migrate to the net namespace of <pid> > >Wouldn't it be more useful to accept an arbitrary path, and just hand it >to setns(2)? That way, the utility would automatically gain support for >new namespace types (for example `setns /proc/$pid/ns/mount`), and would >allow joining namespaces that are linked outside /proc. > >For example, I seem to remember someone saying that `mount --bind >/proc/$pid/ns/net /tmp/foo-namespace` can be used to make that >particular namespace persist even after the process quits. Then `setns >/tmp/foo-namespace someprogram` could be used to join that namespace again. Yeah, that makes sense, I'll respin this tomorrow, and repost. I'm also going to batch the namespace switches, by opening all the file descriptors first, then calling setns on them. It was pointed out to me by doing them one by one like this, if you setns the mount namespace you might loose access to the paths that were specified. Thanks! Neil -- To unsubscribe from this list: send the line "unsubscribe util-linux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
